In addition to bias, there’s also just the basic problem of knowing what facts are.
An AI wouldn’t be able to tell how nonsensical a proposal for gene therapy targeting the nuclei of diseases mature red blood cells would be.
Unless you know biology, that sounds pretty cool.
AI knows more biology than you think. Gemini said "A proposal for gene therapy directly targeting the nuclei of mature red blood cells for disease treatment would face a fundamental and insurmountable challenge: mature red blood cells in mammals do not have a nucleus. [...]"
Well that is heartening! Thanks for doing the test!
It’s still getting too many lizard facts wrong for me to be comfortable with it assessing the quality of a lizard-related research proposal. Or at least the google AI is. I haven’t tried other AIs.
I organize grant proposal review committees and
